Download Single-breathhold, four-dimensional, quantitative assessment of LV

JOURNAL OF MAGNETIC RESONANCE IMAGING 22:59 – 66 (2005)
Original Research
Single-Breathhold, Four-Dimensional, Quantitative
Assessment of LV and RV Function Using
Triggered, Real-Time, Steady-State Free Precession
MRI in Heart Failure Patients
Girish Narayan, MD,1* Krishna Nayak, PhD,2 John Pauly, PhD,3 and Bob Hu, MD2,4
Purpose: To validate a novel, real-time, steady-state free
precession (SSFP), single-breathhold technique for the assessment of left ventricular (LV) and right ventricular (RV)
function in heart failure patients.
Conclusion: In patients with heart failure, real-time, spiral
SSFP allows rapid and accurate assessment of RV and LV
function in a single-breath hold. Using the same strategy,
increased temporal resolution will allow real-time assessment of cardiac wall motion during stress studies.
Materials and Methods: A total of 20 heart failure patients
(mean age 59 ⫾ 17 years) underwent scanning with our
new, real-time, spiral SSFP sequence in which each cardiac
phase was acquired in 118 msec at a resolution of 1.8 ⫻ 1.8
mm. Each cardiac slice (1-cm thick) was automatically advanced based on a cardiac trigger, allowing complete coverage of the heart in a single breathhold. The patients also
underwent LV and RV assessment with the gold standard:
multiple breathhold, cardiac-gated, segmented k-space
strategy. LV and RV end-systolic volume (ESV) and enddiastolic volume (EDV) and LV mass were compared between the two imaging techniques.
Key Words: MRI; imaging; heart failure; SSFP; spirals
J. Magn. Reson. Imaging 2005;22:59 – 66.
© 2005 Wiley-Liss, Inc.
ACCURATE AND REPRODUCIBLE assessment of cardiac function and chamber volume is vital to the care of
heart failure patients (1). However, exact ventricular
volumes and mass are not routinely obtained even in
settings in which quantitative data is useful, partly due
to the measurement inaccuracy of established modalities such as echocardiography (2,3). While previous authors have validated the accuracy of cardiac MRI– derived volume and functional data (4), existing strategies
suffer from measurement variability and decreased
clinical robustness in heart failure patients (5), Current
techniques require ill, dyspneic patients to undertake
multiple, prolonged breathholds in the exact same diaphragmatic position. They also collect and assemble
data over multiple cardiac cycles, causing image degradation during arrhythmias. In addition, the process of
complete volumetric imaging of the entire right ventricle
(RV) and left ventricle (LV) can take as long as 10 –15
minutes, during which time patient motion will compromise image quality and quantitative accuracy. Current techniques are thus ill-suited for routine application in heart failure patients and have hampered the
clinical utility of cardiac MRI.
Recently, steady-state free precession (SSFP) sequences have been introduced to cardiac MR imaging
(6). While its increased signal to noise ratio and bloodmyocardial contrast has improved image quality, the
quantitation of cardiac function continues to suffer
from prolonged image acquisition times and the need
for multiple breathholds. Its incorporation in faster,
real-time imaging strategies has been hampered by the
relative inefficiency of k-space sampling techniques,
with resultant poor temporal and spatial resolution
(7,8). Some investigators have advocated combining real-time techniques with a free-breathing strategy to as-
Results: The new real-time strategy was highly concordant
with the gold standard technique in the assessment of
LVEDV (r ⫽ 0.98), LVESV (r ⫽ 0.98), RVESV (r ⫽ 0.86),
RVEDV (r ⫽ 0.91), LVMASS (r ⫽ 0.95), RVEF (r ⫽ 0.70), and
LVEF (r ⫽ 0.94). The mean bias (95% confidence interval
[CI]) for each parameter is LVEDV: 10.6 cc (cm3) (3.8 –17.4
cc), LVESV: – 0.8 cc (–5.3 to 3.7 cc), RVEDV: 3.7 cc (–5.6 to
13.2 cc), RVESV: –3.1 cc (–11.1 to 4.9 cc), LVMASS: 26 g
(12.4 –39.8 g), RVEF: –2.9% (1.3 to –7.2 %), LVEF: 1.9% (5
to –1.1%). In addition, data acquisition was only nine ⫾ two
seconds with the real-time strategy vs. 312 ⫾ 41 seconds
for the standard technique.
1
Division of Cardiovascular Medicine, Stanford University Hospital,
Stanford, California, USA.
2
Department of Electrical Engineering, University of Southern California, California, USA.
3
Department of Electrical Engineering, Stanford University, Stanford,
California, USA.
4
Division of Cardiology, Palo Alto Medical Foundation, Palo Alto, California, USA.
Contract grant sponsor: National Institutes of Health; Contract grant
number: R01HL067161.
Presented in part at the ISMRM meeting, Kyoto, Japan, 2004, for 2nd
Place Poster Award.
Supplementary material for this article can be found on the JMRI
website at: www.interscience.wiley.com/jpages/1053-1807/suppmat/
index.html.
*Address reprint requests to: G.N., 300 Pasteur Drive, CVRB, Stanford,
CA 94305. E-mail: [email protected]
Received July 29, 2004; Accepted March 31, 2005.
DOI 10.1002/jmri.20358
Published online in Wiley InterScience (www.interscience.wiley.com).
© 2005 Wiley-Liss, Inc.
59
60
Narayan et al.
Table 1
Sequence Comparison
Basis
Standard SSFP
Real-time SSFP
k-space coverage
Slice advancement
FOV
Spatial resolution
Temporal resolution
TR/TE
ST/SP
Flip angle
Other
Segmented, Cartesian 192 ⫻ 192 matrix
Multiple breathhold
35 ⫻ 35 cm
1.8 ⫻ 1.8 mm
60 msec/segmenta
3.7 msec/1.6 msec
10/0 mm
40 degrees
Spiral, 20 interleaves
Cardiac triggered
20 ⫻ 20 cm
1.8 ⫻ 1.8 mm
118 msec/slice/phase (42 msec)b
5.9 msec/2.2 msec
10/0 mm
40 degrees
Adjustable phasec
a
Using view-sharing.
Intermediate temporal phase was displayed using sliding window reconstruction.
c
Phase at the flip angle could be interactively adjusted to move banding artifacts outside the region of interest.
b
sess LV and RV volumes quantitatively. However, freebreathing strategies are prone to slice misregistration
errors and may compromise interstudy reproducibility.
We have recently described a new high-resolution spiral
sequence that provides efficient k-space sampling for
real-time imaging with SSFP (9). This strategy provides
complete coverage of RV and LV function in a single
breathhold by employing a multislice strategy in which
one slice is imaged every heartbeat (10). This singlebreathhold strategy was assessed in 20 patients with
congestive heart failure and validated with the gold
standard: gated, segmented k-space, multiple breathhold acquisition.
MATERIALS AND METHODS
Pulse Sequences
Triggered, Real-Time, SSFP Sequence
The standard and comparison sequences are summarized in Table 1. A full technical description of the realtime sequence is provided elsewhere (9). The main details of the pulse sequence is shown in Fig. 1. A realtime, refocused, spiral SSFP sequence was implemented
with a 640 ␮s slice-selective excitation pulse followed
by 2.4 msec spiral readouts with M0 and M1 (zero and
Figure 1. Diagram of spiral, refocused, SSFP sequence; 1.2
msec excitation (640 ␮s of RF), 2.4 msec readout, 1.4 msec
rewinder (for zero and first moment).
first moment) refocusing gradients, and a 1.4-msec rewinder. Using an imaging TR of 5.9 msec and 20 interleaves, spatial resolution of 1.8 ⫻ 1.8 mm was achieved
over a 20-cm field of view (FOV) every 118 msec. Sliding
window reconstruction was used to display intermediate temporal phase at 24 frames/second (every 42
msec). Slice thickness was 10 mm with no interslice
gap. As SSFP is highly sensitive to off-resonance, radiofrequency (RF) phase cycling could be interactively adjusted using the real-time user interface to place SSFP
banding outside the region of interest (9).
Standard SSFP Sequence
A cardiac-gated, segmented k-space, SSFP sequence
((FIESTA), GE Medical Systems, Milwaukee, WI, USA) was
used as the standard comparison sequence. Sequence
details are provided in Table 1. Multiple slices were prescribed from base to apex. Each slice was acquired consecutively during a 15–20-second breathhold.
Imaging Protocol
All studies were performed in a GE Signa 1.5-T scanner
(GE Medical Systems, Milwaukee, WI, USA) equipped
with the Stanford real-time interactive (RTI) system (11)
The scanner was equipped with gradients supporting a
magnitude of 40 mT/m and slew rate of 150 T/m/
second. A five-inch surface coil was used as a receiver
in all studies.
A total of 20 patients (mean age 59 ⫾ 17 years, 13
men, and seven women) with a history of clinical heart
failure were recruited from heart failure clinics. Informed consent was obtained for all subjects in accordance with our institutional review board. The primary
diagnoses of the patients were idiopathic dilated cardiomyopathy (N ⫽ 4), ischemic heart disease (N ⫽ 7), moderate to severe mitral regurgitation (N ⫽ 2), primary
pulmonary hypertension (N ⫽ 2), pulmonary regurgitation (N ⫽ 1), hypertrophic cardiomyopathy (N ⫽ 1),
combined pulmonary and aortic insufficiency (N ⫽ 1),
and hypertensive disease (N ⫽ 2). Each underwent
scanning with the standard and new (triggered, realtime spiral SSFP) sequence using the protocol described below.
The standard sequence was used to obtain multiple,
consecutive slices from the base to the apex of the left
Rapid Cardiac Function Assessment With MRI
Figure 2. Desired selection of the trigger delay to delineate the
end-diastolic and end-systolic phases, while avoiding the
steady-state transition.
ventricle. After the standard sequence was completed,
the real-time sequence was initiated. Short-axis views
were localized in real-time. Phase cycling was adjusted
to optimize image quality. The exact position of the
basal slice obtained on the standard sequence was localized in real-time, with fine adjustments made during
the breath-hold. After the requested breathhold, the
sequence was switched to a “triggered” mode, in which
each slice was acquired and automatically advanced
based on a cardiac trigger. The trigger delay was adjusted to ensure adequate delineation of the end-diastolic and end-systolic phases of the cardiac cycle (see
Fig. 2). In addition, slice advancement was halted for
short RR intervals, (defined as less than 75% of the
basal RR interval). This prevented slice advancement
after premature beats. All acquisition was performed
during continuous scanning. The total number of slices
was adjusted based on the size of the heart. Generally,
8 –12 slices could be obtained in a breathhold lasting
8 –12 RR intervals. Each time a slice location was
shifted, remaining magnetization of the old slice was
spoiled. For both sequences, breathholding was performed in end-expiration. Both sequences were performed on the same day.
Image Analysis
To determine the accuracy of RV and LV volume assessment, we calculated parameters for both techniques.
Manual segmentation of the LV and RV end-diastolic
and end-systolic endocardial surfaces and LV end-diastolic epicardial surface was completed off-line using
the MASS software package (Medis Inc., Leiden, the
Netherlands) for the standard sequence images and
Scion Image (Scion Corp., Frederick, MD, USA) for the
real-time images. Two separate packages were used, as
neither package could handle both data sources. Calculation of volumes of phantoms imaged with both
techniques demonstrates excellent agreement when
analyzed by the two packages (data not shown). The
papillary muscles were not included in the analysis.
The basal section of the LV was defined as the slice at
which at least 50% of the LV myocardial circumference
was visible in all cardiac phases. The basal diastolic
and systolic slice of the RV was the first slice not to
include any part of the RV outflow tract. The exact
starting slice position used in the standard sequence
was obtained by real-time adjustment of the plane
while using the new, comparison sequence. This ensured similar basal slice positioning between the two
sequences and minimized error resulting from disparate slice correspondence between the two sequences.
61
End-diastolic volume (EDV) and end-systolic volume
(ESV) were calculated by summing the area determined
in each slice multiplied by the slice thickness. LV mass
(LVMASS) was calculated using the difference between
the end-diastolic epicardial versus endocardial volumes
multiplied by the density of the myocardium (1.05
g/cm3). Datasets were reviewed after a period of at least
two weeks by the same individual and by a different
individual experienced in cardiac MR to obtain intraand interobserver variability.
Statistical Evaluation
Quantitative values of LVEDV, LVESV, RVEDV,
RVESV, and LVMASS were compared between the standard technique and the triggered, real-time, SSFP technique using correlation analysis. Bland-Altman (12)
analysis was performed to assess for any systematic
differences between the two techniques. Intra- and interobserver variability was calculated as the percentage
of the absolute difference between the measurements
divided by the mean of the two measurements.
RESULTS
Accuracy of Volumetric Measurements
Representative movie clips and still frames of the shortaxis view obtained using our triggered, real-time, SSFP
sequence in a patient with dilated cardiomyopathy and
in a patient with RV enlargement demonstrate excellent
blood-myocardial definition (Fig. 3; and Supplementary
Movie (Supplementary material for this article can be
found on the JMRI website at: www.interscience.wiley.com/jpages/1053-1807/suppmat/index.html.). In all
studies, satisfactory end-diastolic and end-systolic
frames were obtained at each slice location to allow
accurate border delineation for subsequent volume calculation. In the real-time images, upon each slice shift,
there was a noticeable period during which the new
slice reached steady state. This did not interfere with
the assessment of function or volume as the trigger
delay was adjusted to place the relevant end-diastolic
and end-systolic frames outside this transient period.
There was excellent agreement between the standard
sequence and the triggered, real-time, SSFP approach
for volumetric and functional assessments (see Fig. 4;
left). The results of the Bland-Altman analysis are
shown in Fig. 4 (right). The assessments of LVEDV and
LVMASS obtained with the triggered, real-time approach were 10.6 ⫾ 3.2 cc (cm3) and 26 ⫾ 6.8 g (mean ⫾
SD) higher than those obtained using the standard sequence (P ⬍ 0.05). However, the other parameters
showed no statistically significant evidence of systematic difference. Inter- and intraobserver variability are
presented in Table 2.
Time Efficiency
All patients were able to complete the required protocol.
The mean scan time for complete volumetric coverage of
the heart after short axis localization was nine ⫾ two
seconds for the real-time vs. 312 ⫾ 41 seconds for the
standard approach. The latter time included recovery
62
Figure 3. End-diastolic (left) and end-systolic (right) frames
from a patient with dilated cardiomyopathy acquired in a single breathhold from base to apex using the new triggered,
real-time, SSFP sequence.
time between breathholds as deemed necessary by the
patient. Heart rates ranged from 53 to 93 beats per
minute, resulting in a single breathhold duration for
the real-time approach of seven to 15 seconds as compared to eight to 12 breathholds of 15 to 20 seconds
each for the standard sequence.
DISCUSSION
We have demonstrated that, in a single breathhold, LV
and RV function and volumes can be accurately quantified in heart failure patients with a real-time, spiral
SSFP sequence.
Currently, the common clinical approach for this
purpose includes CINE MR techniques, which are
based on cardiac gated, multiphase, segmented
k-space sequences, in which every slice acquisition requires a breathhold (13–15). However, the requirement
for multiple breathholds and the need to combine data
from multiple cardiac cycles complicates clinical scans
Narayan et al.
in patients with dyspnea and arrhythmias, characteristics present in the heart failure population (1). In
addition, these strategies, as currently implemented,
require a prescription phase during which scan planes
are identified, followed after a length of time by the
actual scans. This strategy has the potential to introduce patient motion related positioning errors and also
extends overall scan time. As the study represents only
a part of the cardiac examination, patients often complain of excessive overall study duration.
In order to address these limitations, real-time techniques, in which cardiac motion is captured without the
need for gating, allows an interactive, rapid cardiac
examination and has proven useful for the assessment
of ventricular volumes (11,16 –19,20). However, these
previous real-time techniques were based on gradient
recalled echo sequences, which primarily rely on T1
and in-flow to produce contrast. In comparison, SSFP
techniques produce enhanced endocardial-blood contrast as a result of its T2 and T1 dependence and effective use of all available signal. As a result, it has been
widely adopted for rapid cine imaging of the heart.
We have combined the inherently high blood-myocardial contrast characteristics of SSFP sequences with
the advantages of the real-time approach. Other authors have also described the use of SSFP for real-time
imaging (7,8). However, Hori et al (7), employed the
relatively less efficient two-dimensional Fourier transform (FT) strategy with a resulting true temporal resolution of 164 msec and a spatial resolution of 4.17 ⫻
2.73 mm. Spuentrup et al (8) achieved a higher spatial
resolution of 2.5 ⫻ 2.5 mm, though at a much lower
true temporal resolution of 200 msec. In addition, their
use of projection reconstruction grossly undersamples
k-space data. In comparison, using spiral k-space trajectories, we have achieved a true temporal resolution
of 118 msec and 1.8 ⫻ 1.8 mm resolution without any
undersampling. The improved temporal and spatial
resolution will better delineate cardiac borders, especially during the rapidly occurring systolic phase (19).
Temporal resolutions of approximately 90 msec have
previously been shown to demonstrate good volumetric
assessment during end-systole and end-diastole. Our
temporal resolution (118 msec) is close to this value.
Furthermore, Foo et al (21) demonstrated that the use
of the intermediate temporal phase is important in improving the temporal characteristics of the sequence.
The use of the sliding window reconstruction technique, in our technique, provides 24 images per second.
Finally, the performance of the sequence demonstrates
close agreement of the quantitative values to the gold
standard values obtained in our study.
Overall, our technique yielded very accurate and reproducible values for LV and RV volumes as compared
to the standard technique. In this study, the LVEDV
and LVMASS were larger than the values determined by
the standard technique by 6.6% (P ⬍ 0.05) and 13%
(P ⬍ 0.05), respectively. In addition, the RVEDV showed
a trend toward being larger (P ⬍ 0.07) as compared to
the gold standard acquisition. This leads us to speculate that the multiple breathhold sequence may introduce systematic errors due to changes in expiratory
position and resulting slice misregistration.
Rapid Cardiac Function Assessment With MRI
63
Figure 4. Correlation of FIESTA and triggered, real-time (left) and bias plot (right). The correlation coefficients are LVESV: r ⫽
0.98; LVEDV: r ⫽ 0.98; RVESV: r ⫽ 0.86; RVEDV: r ⫽ 0.91; LVMASS: r ⫽ 0.95, RVEF: r ⫽ 0.70; and LVEF: r ⫽ 0.94. The mean
bias is represented by the dotted line. The dashed lines indicate the 95% limits of agreement. The solid line indicates zero bias.
The mean bias (95% CI) for each parameter is LVEDV: 10.6 cc (3.8 –17.4 cc); LVESV: – 0.8 cc (–5.3 to 3.7 cc); RVEDV: 3.7 cc (–5.6
to 13.2 cc); RVESV: –3.1 cc (–11.1 to 4.9 cc); LVMASS: 26 g (12.4 –39.8 g); RVEF: –2.9% (1.3 to –7.2 %); and LVEF: 1.9% (5 to
–1.1%).
The use of the SSFP sequence also introduces transient
artifacts, which result from steady-state discontinuity
when a new slice is imaged. In theory, this may interfere
with the interpretation of cardiac function. However, in
patients, we have found that the duration of this artifact is
small. Furthermore, the trigger delay is optimally adjusted to prevent this transient artifact from interfering
with the end-diastolic phase. Further shortening of the
transients may be possible with advanced steady-state
methods as described by Hargreaves et al (22).
In addition, our values for inter- and intraobserver
variability are similar to those described in the literature for assessments of diastolic and systolic LV and RV
volumes (7,17,19). The values between the standard
and the real-time approach are also quite similar, suggesting that both techniques offer adequate image quality to render reproducible interpretations.
Overall scan time is also decreased using our strategy. The use of a triggered, spiral, real-time approach
allows complete volumetric coverage of the right and left
ventricles in a single breathhold. This approach also
obviates the need for a prolonged localizer step, with
interactive scanning and position occurring in realtime. This dramatically reduces scan time (by an average of about five minutes in our study, not including
scout image acquisition used in conventional techniques) compared to the standard sequence. In addition, patient motion does not require repeating the localization and prescription scans.
Our strategy also decreases the overall breathhold
requirements for patients. We have also coupled realtime techniques with free breathing strategies (17).
However, these require exquisite correlation with diaphragm position in order to minimize slice misregistra-
64
Narayan et al.
Figure 4 (Continued)
tion given respiratory related changes in cardiac position and ventricular volume. As a result, these
techniques may suffer from poor interstudy reproducibility, depend heavily on operator expertise, and thus
lack clinical robustness compared to breathheld scans.
Furthermore, cardiac MR has been proposed as the
technique of choice to assess treatment effects in clinical studies given its accuracy, low interobserver variability, and subsequent potential to dramatically reduce sample size (23). However, the changes in ejection
fraction observed in clinical studies, often on the order
of 8% to 10% (24), require a technique that has low
interstudy variability as well (5). By adopting a singlebreathhold strategy, one can eliminate possible error
introduced by slice-to-slice variation in diaphragm position and resulting slice misregistration. This also
serves to decrease operator and patient dependence, at
the same time minimizing patient breathholds. Given
that most patients, including those with heart failure,
can hold their breath once for at least 10 seconds, a
single breathhold strategy promises to provide clinically
more robust results. Further assessment of interstudy
variability using this technique in a larger group of
patients would quantitate this benefit.
Arrhythmias are also frequently present in the heart
failure population (i.e., ventricular bigeminy, premature ventricular contradiction (PVCs), atrial fibrillation,
etc.) During a segmented k-space approach, the presence of these arrhythmias will frustrate efficient, highquality image acquisition. While collection of this data
over multiple heartbeats may serve to “average” data
over multiple samples, true real-time wall motion is not
depicted. Furthermore, it remains unclear whether the
accuracy of this approach will be superior to averaging
multiple single-volumetric acquisitions using our strategy. Taken to the extreme, standard techniques of “averaging” over multiple cardiac cycles could lead to a
“static” movie of the heart by averaging many possible
positions of the myocardium. In our strategy, during
real-time imaging, each phase is visualized in real-time,
Rapid Cardiac Function Assessment With MRI
65
Figure 4 (Continued
removing the requirement for a regular rhythm. For
volumetric calculations, rejection of short RR intervals,
where the slice was not advanced, may capture a slice
during postextrasystolic augmentation, thus not reflecting true basal LV function. However, the error introduced by this is confined to that one slice. Furthermore, in patients with frequent ectopy, multiple
acquisitions of the entire cardiac volume can be ob-
tained very quickly, allowing one to average over multiple samples. Nonetheless, further study will be needed
in this regard to quantify the benefit of this strategy in
a larger cohort of patients with arrhythmias.
In this study, primary emphasis was placed on the
assessment of ventricular volumes. Though we believe
that accurate wall motion analysis is also obtained in
an analogous rapid fashion, a larger patient cohort with
66
Narayan et al.
Table 2
Intra- and Interobserver Variability*
Intraobserver variability
Triggered
Fiesta
Interobserver variability
Triggered
Fiesta
LVEDV
LVESV
RVEDV
RVESV
LVMASS
5.6 ⫾ 4.5
5.8 ⫾ 4.1
9.2 ⫾ 5.3
8.8 ⫾ 6.2
7.9 ⫾ 3.9
10.1 ⫾ 5.7
7.4 ⫾ 4.1
12.1 ⫾ 4.9
10.5 ⫾ 2.1
8.5 ⫾ 4.9
5.3 ⫾ 2.9
4.4 ⫾ 3.9
15.0 ⫾ 2.4
13.5 ⫾ 8.2
4.5 ⫾ 0.8
8.4 ⫾ 7.0
8.2 ⫾ 9.5
16.3 ⫾ 9.4
13 ⫾ 6.5
10.8 ⫾ 8.5
*All values are expressed as percentages, calculated as the absolute difference of the two measurements divided by the mean ⫾ SD.
a larger sample of regional wall motion abnormalities
would be required to rigorously address this. In this
regard, in contrast to segmented strategies, our strategy allows true depiction of wall motion in real-time.
Parameter adjustment in our strategy to yield improved
temporal resolution (at the expense of spatial resolution) promises to facilitate true real-time myocardial
stress evaluation (25). In conclusion, triggered, realtime, spiral SSFP allows rapid and accurate quantitation of RV and LV function in patients with heart failure. This technique provides a clinically robust, rapid
modality to accurately assess cardiac function in heart
failure patients using cardiac MRI.
REFERENCES
1. Hunt SA, Baker DW, Chin MH, et al. ACC/AHA guidelines for the
evaluation and management of chronic heart failure in the adult: a
report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise
the 1995 Guidelines for the Evaluation and Management of Heart
Failure). 2001. American College of Cardiology Web site. Available
at: http://www.acc.org/clinical/guidelines/failure/hf_index.htm.
2. Bottini PB, Carr AA, Prisant LM, et al. Magnetic resonance imaging
compared to echocardiography to assess left ventricular mass in
the hypertensive patient. Am J Hypertens 1995;8:221–228.
3. Teichholz LE, Kreulen T, Herman MV, et al. Problems in echocardiographic volume determinations: echocardiographic-angiographic correlations in the presence or absence of asynergy. Am J
Cardiol 1976;37:7–11.
4. Bellenger NG, Grothues F, Smith G, et al. Quantification of right
and left ventricular function with MRI. Herz 2000;25:392–399.
5. Grothues F, Smith GC, Moon JC, et al., Comparison of interstudy
reproducibility of cardiovascular magnetic resonance with two-dimensional echocardiography in normal subjects and in patients with heart
failure or left ventricular hypertrophy. Am J Cardiol 2002;90:29 –34.
6. Thiele H, Nagel E, Paetsch I, et al. Functional cardiac MR imaging
with steady-state free precession (SSFP) significantly improves endocardial border delineation without contrast agents. J Magn Reson Imaging 2001;14:362–367.
7. Hori Y, Yamada N, Higashi M, et al. Rapid evaluation of right and
left ventricular function and mass using real-time true-FISP cine
MR imaging without breath-hold: comparison with segmented trueFISP cine MR imaging with breath-hold, J Cardiovasc Magn Reson
2003;5:439 – 450.
8. Spuentrup E, Schroeder J, Mahnken AH, et al. Quantitative assessment of left ventricular function with interactive real-time spiral
and radial MR imaging. Radiology 2003;227:870 – 876.
9. Nayak KS, Hargreaves B, Hu BS, et al., Spiral balanced SSFP
cardiac imaging. Magn Reson Med 53:1468 –1473.
10. Nayak K, Hu BS. Triggered real-time MRI and cardiac applications.
Magn Reson Med 2003;49:188 –192.
11. Kerr ABS, Li KCP, Handy CJ, Mayer CH, Macovski A, Nishimura
DG. Real-time interactive MRI on a conventional scanner. Magn
Reson Med 1997;38:355–367.
12. Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986;1:307–
310.
13. Buser PT, Aufermann W, Holt WW, et al. Noninvasive evaluation of
global left ventricular function with use of cine nuclear magnetic
resonance. J Am Coll Cardiol 1989;13:1294 –1300.
14. Sakuma H, Fujita N, Foo TK, et al. Evaluation of left ventricular
volume and mass with breath-hold cine MR imaging. Radiology
1993;188:377–380.
15. Sechtem U, Pfugfelder PW, Gould RG, et al. Measurement of right
and left ventricular volumes in healthy individuals with cine MR
imaging. Radiology 1987;163:697–702.
16. Yang PC, Karr AB, Liu AC, et al. New real-time interactive magnetic
resonance imaging complements echocardiography. J Am Coll Cardiol 1998;32:2049 –2056.
17. Kaji S, Yang PC, Kerr AB, et al. Rapid evaluation of left ventricular
volume and mass without breath-holding using real-time interactive cardiac magnetic resonance imaging system. J Am Coll Cardiol
2001;38:527–533.
18. Nagel E, Schnieder U, Ibrahim T, et al. Magnetic resonance realtime imaging for the evaluation of left ventricular function. J Cardiovasc Magn Reson 2000;2:7–14.
19. Setser RM, Fischer SE, Lorenz CH. Quantification of left ventricular
function with magnetic resonance images acquired in real time. J
Magn Reson Imaging 2000;12:430 – 438.
20. Bornstedt A, Nagel E, Schalla S, et al. Multi-slice dynamic imaging:
complete functional cardiac MR examination within 15 seconds. J
Magn Reson Imaging 2001;14:300 –305.
21. Foo TK, Bernstein MA, Aisen AM, et al. Improved ejection fraction
and flow velocity estimates with use of view-sharing and uniform
repetition time excitation with fast cardiac techniques. Radiology
1995;195:471– 478.
22. Hargreaves BA, Vasanawalla SS, Pauly JM, et al. Characterization
and reduction of the transient stead-state in MR imaging. Magn
Reson Med 2001;46:149 –153.
23. Bellenger NG, Davies LC, Francis JM, et al. Reduction in sample
size for studies of remodeling in heart failure by the use of
cardiovascular magnetic resonance J Cardiovasc Magn Reson
2000;2;271–278.
24. Packer M, Antonopoulos GV, Berlin JA, et al. Comparative effects of
carvedilol and metoprolol on left ventricular ejection fraction in
heart failure: results of a meta-analysis. Am Heart J 2001;141:
899 –907.
25. Nayak KS, Pauly J, Nishimura DG, et al. Rapid ventricular assessment using real-time interactive multislice MRI. Magn Reson Med
2001;45:371–375.